We propose to conduct a molecular epidemiologic study of sporadic retinoblastoma (RB), when it occurs without a family history of the disease. RB, a cancer of the embryonal retina in infants and young children, results from mutation in the RB1 gene and can be bilateral or unilateral. In bilateral RB, the critical mutation occurs almost always on the father's gamete before the child's conception. In unilateral RB, mutation occurs after the child's conception, that is during gestation or early postnatal life. Our model for sporadic RB proposes a role for parental genotypes of carcinogen metabolizing enzymes (CME), DMA repair genes and exposures in determining the risk for a mutation in RB1 gene. The genotype for CME of an individual can influence the level and duration of exposure to a putative carcinogen and the resultant DMA damages. Similarly, the DMA repair genotypes define the efficiency of damage removal, and if not repaired, damages lead to mutations. If the mutation occurs in RB1 gene in a sperm precursor (sporadic bilateral RB) or a developing retinal precursor cell (unilateral RB), retinoblastoma results. For bilateral RB, we hypothesize that polymorphisms in the genes of the father with negative functional consequences increase risk, as do his occupational, dietary, x-ray, tobacco, and alcohol exposures before the child's conception. For unilateral RB, we hypothesize that the polymorphisms in the genes carried by the mother and the child and the exposures during the pregnancy increase risk. The effect of chemical and physical exposures can be specific in that they cause particular types of DNA damages that, if not repaired, lead to particular types of mutations. The mutations in RB1 gene that result in RB can be detected and characterized in a large number of cases. Therefore, we propose to investigate the relationship between specific CMEs, DNA repair pathways, specific exposures, and specific types of RB1 mutations. Cases of unilateral and bilateral RB will be ascertained through the participating hospitals of the Children's Oncology Group, six additional participating centers and Will's Eye Hospital - centers that treat most children with RB in the U.S. Controls will be ascertained through the birth certificates. Case-control comparisons will be made to test hypotheses about polymorphisms in CME and DNA repair genes, and exposures for bilateral and unilateral RB. To test hypotheses about subsets of cases defined by type of RB1 mutation, we will use case-case comparisons. Much has been learned about the mechanism and genetics of cancer from the study of retinoblastoma. We believe the usefulness of retinoblastoma as a paradigm extends to the role of genes other than the disease gene and environmental exposures in childhood and adult cancer.